testing-llm
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides best practices for testing AI features and does not contain any malicious code, command execution, or dangerous instructions.
- [DATA_EXFILTRATION]: No data exfiltration patterns detected. The instructions specifically advise users to filter sensitive headers (e.g., 'authorization', 'x-api-key') when using VCR.py for integration tests, which is a positive security practice.
- [REMOTE_CODE_EXECUTION]: No remote code execution patterns or untrusted downloads were found. All referenced libraries (DeepEval, RAGAS, VCR.py) are well-known industry standards.
- [PROMPT_INJECTION]: No prompt injection attempts or bypass patterns were found in the skill's instructions. The checklist explicitly suggests testing the target application for prompt injection resistance.
- [OBFUSCATION]: No obfuscated content, hidden strings, or suspicious encoding techniques were found in the files.
Audit Metadata